[llvm] [AMDGPU] Add scaffolding for ML focused scheduling strategy (PR #169616)

[Shilei Tian via llvm-commits]

================
@@ -331,6 +331,14 @@ void GCNSubtarget::overrideSchedPolicy(MachineSchedPolicy &Policy,
   Policy.OnlyTopDown = false;
   Policy.OnlyBottomUp = false;
 
+  const Function &F = Region.RegionBegin->getMF()->getFunction();
+  Attribute WorkloadAttr = F.getFnAttribute("amdgpu-workload-type");
+  bool IsMLWorkload =
+      WorkloadAttr.isValid() && WorkloadAttr.getValueAsString() == "ml";
+  // Always schedule top-down for better blancing of HW resource usage.
+  if (IsMLWorkload)
----------------
shiltian wrote:

How is this gonna interact with front end?

https://github.com/llvm/llvm-project/pull/169616